<aside> 🗣️ Goal for today: Remove Markov assumption when build model.

</aside>

Overview

Recurrent networks for language modeling

Long Short-term Memory (LSTM)

Some additional informations

Recurrent Neural Network (RNN) for language Modeling

background:

From Feedforward to Recurrent Networks

Training

Vanishing Gradient problem

Long Short-term Memory (LSTM)

Architecture:

Training:

Variant - The Gated Recurrent Unit (GRU)